Modelling soil CO2 production and transport with dynamic source and diffusion terms: Testing the steady-state assumption using DETECT v1.0

Ryan, Edmund and Ogle, Kiona and Kropp, Heather and Samuels-Crow, Kimberley and Carrillo, Yolima and Pendall, Elise (2018) Modelling soil CO2 production and transport with dynamic source and diffusion terms: Testing the steady-state assumption using DETECT v1.0. Geoscientific Model Development, 11 (5). pp. 1909-1928. ISSN 1991-959X

Abstract

The flux of CO2 from the soil to the atmosphere (soil respiration, Rsoil) is a major component of the global carbon cycle. Methods to measure and model Rsoil, or partition it into different components, often rely on the assumption that soil CO2 concentrations and fluxes are in steady state, implying that Rsoil is equal to the rate at which CO2 is produced by soil microbial and root respiration. Recent research, however, questions the validity of this assumption. Thus, the aim of this work was two-fold: (1) to describe a non-steady state (NSS) soil CO2 transport and production model, DETECT, and (2) to use this model to evaluate the environmental conditions under which Rsoil and CO2 production are likely in NSS. The backbone of DETECT is a non-homogeneous, partial differential equation (PDE) that describes production and transport of soil CO2, which we solve numerically at fine spatial and temporal resolution (e.g., 0.01 m increments to 1 m, every 6 hours). Production of soil CO2 is simulated for every depth and time increment as the sum of root respiration and microbial decomposition of soil organic matter, both of which can be driven by current and antecedent soil water content and temperature, which can also vary by time and depth. We also analytically solved the ordinary differential equation (ODE) corresponding to the steady-state (SS) solution to the PDE model. We applied the DETECT NSS and SS models to the 6-month growing season period representative of a native grassland in Wyoming. Simulation experiments were conducted with both model versions to evaluate factors that could affect departure from SS: (1) varying soil texture; (2) shifting the timing or frequency of precipitation; and (3) with and without the environmental antecedent drivers. For a coarse-textured soil, Rsoil from the SS model closely matched that of the NSS model. However, in a fine-textured (clay) soil, growing season Rsoil was ~3% higher under the assumption of NSS (versus SS). These differences were exaggerated in clay soil at daily time-scales whereby Rsoil under the SS assumption deviated from NSS by up to ~20% in the 10 days following a major precipitation event. Moreover, incorporation of antecedent drivers increased the magnitude of Rsoil by 15% to 37% for coarse- and fine-textured soils, respectively. However, the responses of Rsoil to the timing of precipitation and antecedent drivers did not differ between SS and NSS assumptions. In summary, the assumption of SS conditions can be violated depending on soil type and soil moisture status, as affected by precipitation inputs, and the DETECT model provides a framework for accommodating NSS conditions to better predict Rsoil and associated soil carbon cycling processes.